Modern high power transmitters such as those used for television or for radar transmission require vacuum tubes which operate at relatively high voltages and which draw substantial amounts of current. It is relatively easy to supply direct high voltage at high current by applying alternating current from the power mains to the primary winding of a transformer, the secondary winding of which produces a high alternating voltage. High voltage rectifiers are connected to the second winding of the transformer to rectify the alternating current to produce direct current. Filter capacitors may be used to reduce voltage ripple. Such an arrangement tends to be bulky and expensive, because of the large physical size of the transformer required for operation at 50 or 60 Hertz (Hz) power line frequencies and because of the very large values of filter capacitance and associated working voltage of the requisite filter capacitors.
It is known that operation of a transformer at high frequencies, such as 1000 to 10,000 Hz, can effect a significant reduction in physical size. Furthermore, operation at high frequencies reduces the capacitance of the requisite filter capacitors in proportion to the ratio of frequencies. It is often desirable to operate an inverter at a frequency which is equal to or a multiple of the operating frequency of the system being powered. For example, in television practice, it is common to operate the high voltage kinescope power supply at 15,750 Hz, which is the horizontal deflection frequency.
In order to generate the high frequency drive for the primary of a voltage step-up transformer, it is common to produce a relatively low direct voltage by rectifying and filtering the power mains, and to apply the relatively low direct voltage so produced to a switched bridge inverter in order to generate an alternating potential at a high frequency which may be applied to the primary winding of the step-up transformer. In order to maximize the efficiency of conversion of power from the alternating current power mains into radio frequency (RF) power, it is important to minimize losses in the high frequency inverter. These losses are for the most part associated with the solid state switching elements of the inverter, and include rise time loss, conduction loss and fall time loss. In most practical applications, rise time loss is relatively small, since circuit inductances limit the rise time of the currents. The presence of these inductances causes the voltage across the switching element to decrease to near zero before significant current flows through the switching device. Conduction loss through the solid state switching elements is minimized by proper selection of the switching device itself, by paralleling of the switching devices if necessary, and by application of sufficient switch drive. In the case of bipolar transistor switches, the switch drive is base current, and in the case of FET switches, the switch drive is gate voltage.
In a bipolar transistor inverter driven from a source of substantially constant voltage, the collector voltage of each switch transistor ordinarily rises to substantially its maximum value before the collector current begins to decrease at each turn-off interval. During the period when collector voltage is applied and collector current continues to flow, fall time losses occur. Fall time losses increase linearly as operating frequency is increased, and tend to be the most important of the losses in the inverter. In a bridge type inverter to which a substantially constant voltage is applied, fall time loss is proportional to the product of the peak voltage and current, the cross-over time and the frequency. Cross-over time is the time interval between the beginning of the voltage rise across a switch (commonly measured at the 10% point) until the end of the current fall (10% of the maximum value).
Switch stress and fall time losses can be reduced as described in patent application Ser. No. 850,813, entitled "Current Fed Inverter Bridge With Conduction Overlap And Load Tuning", filed Apr. 11, 1986, in the name of Pruitt, by driving the inverter bridge from a constant current source, and by operating the switches of the inverter in a recurrent sequence of four intervals, during two of which intervals all of the switches of the inverter are conductive, thereby short-circuiting the current source. During the short-circuited intervals, a resonant circuit is set up which causes oscillatory currents to flow through the conductive switches, which currents periodically reduce the net current through the switches in preparation for opening of (rendering nonconductive) the switches. Due to normal circuit tolerances, and also due to reduction in oscillatory current attributable to dissipative losses, or because of transfer of energy from the oscillatory circuit to the load, the oscillatory current may not reduce the current through a switch being opened to zero. A finite current therefore continues to flow through the switch as it is being opened, which results in stress on the switch.
It is desirable to reduce the stress on the switches of an inverter bridge operated in a current fed mode with overlapping conduction periods in a manner which does not dissipate large amounts of energy, and which does not require complex timing control.